Transcript
Intel Technology Briefing
PCI Local Bus A New Era In Speed
Next CPage
C LICK
TO
ONTINUE
PCI: Laying the Tracks for Today’s Technology This Technology Briefing will tell you about a new local bus technology—PCI. Jointly developed by Intel and other industry leaders, it’s designed for current and next-generation PCs. PCI picks up where today’s bus architectures leave off, providing greater system performance via a wider data path and increased expandability.
Next Page
Start Over
INTEL
Main Menu Click on the buttons for more information.
PCI LOCAL BUS
CPU Bus
PCI
The PCI Story Chart on PCI Performance
These are the features of the new PCI local bus technology.
Chip Set 33MHz x 32-Bit = 132 MB /sec
PCI Bus
Burst Mode Data Transfer
High-Speed Buffers
AutoConfiguration NETWORKING
GRAPHICS
VIDEO
ISA Bus and Controller
DISK DRIVES
FUTURE PERIPHERALS
Get More Information
Start Over
PREVIOUS
Page
Burst Mode Data Transfer
P
CI boasts a 32-bit data path
High-Speed Buffers
CPU
and a 33 MHz clock speed.
Bus
This yields a maximum data transfer
PCI Chip Set
PCI Bus
rate of 132 megabytes per second,
AutoConfiguration
33MHz x 32-Bit = 132 MB/sec
a marked improvement over the 5 MB/sec. for a standard ISA bus.
Video
Networking Graphics
Disk Future Drives Peripherals
ISA Bus and Controller
And with PCI, data is written to and read from the peripherals through linear bursts. This means the data is transferred every clock cycle, doubling PCI’s throughput over buses without linear bursting capabilities.
Main Menu
Start Over
Previous Page
P
PCI Buffered Expansion Architecture CI’s architecture features a Bus
CPU and peripherals that amplifies the signal, allowing multiple high-
Burst Mode Data Transfer
CPU
buffer design between the
PCI Chip Set
PCI Bus
AutoConfiguration
33MHz x 32-Bit = 132 MB/sec
speed peripherals to be attached to the same PCI local bus. Buffering
Networking Graphics
also isolates the peripherals from
Video
Disk Future Drives Peripherals
ISA Bus and Controller
the CPU, reducing noise and increasing reliability. And because PCI interfaces with ISA, EISA and Microchannel buses, you can continue to use existing add-in cards.
Main Menu
Start Over
Previous Page
A Jumperless Future
T
hrough PCI, Intel is working Bus
PCI
practical reality. At the heart of PCI’s design, built-in configuration registers
Burst Mode Data Transfer
CPU
to make auto-configuration a
Chip Set
PCI Bus
High-Speed Buffers
33MHz x 32-Bit = 132 MB/sec
and software will automatically keep track of every interrupt being used in
Networking Graphics
Video
Disk Future Drives Peripherals
ISA Bus and Controller
the system. When a new PCI peripheral is added, the PCI chipset will simply select an unused interrupt. No more changing jumpers. No more keeping track of IRQs.
Main Menu
Start Over
Previous Page
CI is the best local bus implementation for the following reasons:
•
30 Megabytes/Sec
P
PCI: Where Local Bus is Going 25 20
Chart Details
15 10
Highest performing bus today. 5 The Torque Benchmark on the right illustrates PCI’s bus throughput for graphics operations. It measures the rate at which bytes are transferred from main memory to screen. Pathway to capabilities like plug and play. Offers the best expandability for high performance peripheral devices.
PCI VESA Bus ISA SVGA SVGA SVGA
• •
ISA VGA
What is a Local Bus? Break the Bottleneck Room to Grow Main Menu
Start Over
Previous Page
Bus Throughput
T
30 Megabytes/Sec
his benchmark was run on identically configured i486™ DX2-66 CPU-based systems. Only the bus architecture, and therefore the graphics card, was changed. All SVGA graphics cards used versions of ATI’s Mach32* graphics accelerator family.
Source: Torque Benchmark.
25 20 15 10 5 PCI VESA Bus ISA SVGA SVGA SVGA
ISA VGA
Main Menu
Start Over
Previous Page
Local Bus in a Nutshell
S
imply put, a local bus takes peripherals off the I/O bus and connects them, together with the CPU and the memory subsystem, to a wider, faster pathway for data. The result is faster data transfer between the CPU and the peripherals. That’s especially important for servers and graphic-intensive software like Windows and OS/2.*
Main Menu
Start Over
Next Page
PREVIOUS
Page
T
How to Break the Bottleneck
he I/O bus most people use today, ISA, was specified for the original 286 computer. Its 16-bit data path and 8 MHz clock speed provided satisfactory performance for moving CGA graphic images of 150 thousand bytes and DOS applications limited to several hundred thousand kilobytes. But with current generation Super VGA images requiring 750 KBytes and Windows* applications weighing in at several megabytes, traditional I/O buses are unable to transfer the information quickly enough. The solution? A local bus system capable of moving 32 bits of data at 33 MHz.
MEGABYTES PER SEC.
Main Menu
Start Over
Next Page
PREVIOUS
Page
Room to Grow
M
ost PCI systems will support three to five performance-critical peripherals. These peripherals will be either integrated directly onto the motherboard or can be added via PCI expansion cards, such as multimedia, graphics, disk drives and LAN cards. Of course, you can still use existing ISA, EISA or Microchannel add-in cards because PCI is designed to supplement, not replace, the traditional I/O bus.
Main Menu
Start Over
Previous Page
T
Want to Learn More?
o find out more about PCI local bus technology, call us and we’ll gladly send you a technical overview with benchmarks, a list of PCI systems and add-in boards, and information on the PCI SIG. Reprints of the Pentium™ processor and OverDrive™ Processor Technology Briefings are also available ® through our toll-free number.
Call
1-800-955-5599 and ask for literature packet #69.
MAIN MENU
Start Over
©1994 Intel Corporation. *All products mentioned are trademarks of their respective companies.
Previous Page
